Mutant p53 regulates enhancer-associated H3K4 monomethylation through interactions with the methyltransferase MLL4

Monomethylation of histone H3 lysine 4 (H3K4me1) is enriched at enhancers that are primed for activation and the levels of this histone mark are frequently altered in various human cancers. Yet, how alterations in H3K4me1 are established and the consequences of these epigenetic changes in tumorigenesis are not well understood. Using ChIP-Seq in human colon cancer cells, we demonstrate that mutant p53 depletion results in decreased H3K4me1 levels at active enhancers that reveal a striking colocalization of mutant p53 and the H3K4 monomethyltransferase MLL4 following chronic tumor necrosis factor alpha (TNFα) signaling. We further reveal that mutant p53 forms physiological associations and direct interactions with MLL4 and promotes the enhancer binding of MLL4, which is required for TNFα-inducible H3K4me1 and histone H3 lysine 27 acetylation (H3K27ac) levels, enhancer-derived transcript (eRNA) synthesis, and mutant p53-dependent target gene activation. Complementary in vitro studies with recombinant chromatin and purified proteins demonstrate that binding of the MLL3/4 complex and H3K4me1 deposition is enhanced by mutant p53 and p300-mediated acetylation, which in turn reflects a MLL3/4-dependent enhancement of mutant p53 and p300-dependent transcriptional activation. Collectively, our findings establish a mechanism in which mutant p53 cooperates with MLL4 to regulate aberrant enhancer activity and tumor-promoting gene expression in response to chronic immune signaling.

Monomethylation of histone H3 lysine 4 (H3K4me1) is enriched at enhancers that are primed for activation and the levels of this histone mark are frequently altered in various human cancers. Yet, how alterations in H3K4me1 are established and the consequences of these epigenetic changes in tumorigenesis are not well understood. Using ChIP-Seq in human colon cancer cells, we demonstrate that mutant p53 depletion results in decreased H3K4me1 levels at active enhancers that reveal a striking colocalization of mutant p53 and the H3K4 monomethyltransferase MLL4 following chronic tumor necrosis factor alpha (TNF␣) signaling. We further reveal that mutant p53 forms physiological associations and direct interactions with MLL4 and promotes the enhancer binding of MLL4, which is required for TNF␣-inducible H3K4me1 and histone H3 lysine 27 acetylation (H3K27ac) levels, enhancer-derived transcript (eRNA) synthesis, and mutant p53-dependent target gene activation. Complementary in vitro studies with recombinant chromatin and purified proteins demonstrate that binding of the MLL3/4 complex and H3K4me1 deposition is enhanced by mutant p53 and p300-mediated acetylation, which in turn reflects a MLL3/4-dependent enhancement of mutant p53 and p300-dependent transcriptional activation. Collectively, our findings establish a mechanism in which mutant p53 cooperates with MLL4 to regulate aberrant enhancer activity and tumorpromoting gene expression in response to chronic immune signaling.
Genome-wide studies have revealed that enhancers are common targets for genetic and epigenetic alterations that are a significant driving force for all types of human cancers (17)(18)(19)(20)(21)(22)(23). Specifically, gains and losses in H3K4me1 accumulation at enhancers have been reported in human cancers (17,24,25) suggesting that alterations in this chromatin modification may lead to enhancer and gene misregulation. In addition, the MLL3 and MLL4 genes are frequently mutated, resulting in various developmental defects and cancers (26 -28), which may result in the misregulation of H3K4me1 deposition and associated enhancer activation. However, the steps required to establish epigenetic alterations at enhancers that are linked to oncogenic gene expression programs remain poorly understood. Of special interest here is the connection between epigenetic alterations and mutations in the tumor suppressor p53, which is the most frequently altered gene found in human cancer (29).
The majority of p53 mutations are missense and are located in p53's conserved DNA binding domain. The resulting mutant p53 proteins reveal an inactivation of p53 tumor suppressor functions and gain-of-function (GOF) activities that contribute to tumor development (30 -32). Increasing evidence has linked GOF properties to the gene regulatory roles of p53 mutants that are largely mediated through mutant p53 recruitment by other transcription factors to promoter and enhancer regions (33)(34)(35)(36)(37)(38). In this regard, mutant p53 has recently been shown to cooperate with the master proinflammatory regulator NF-B to regulate enhancer and gene activation in response to chronic tumor necrosis factor ␣ (TNF␣) signaling with implications for colon cancer aggressiveness (39). However, the mechanisms by which mutant p53 contributes to enhancer regulation and alterations in the cancer cell transcriptome remain to be elucidated. Given WT p53's role in chromatin regulation, p53 mutants may also affect epigenetic changes in chromatin. However, other than the finding that mutant p53 regulates chromatin pathways by modulating the expression of the chromatin modifying enzymes, MLL1 and MLL2 (38), the functional evidence that links chromatin regulation and p53 mutations in tumorigenesis remains elusive.
In the present study, we reveal a novel role for mutant p53 in enhancer regulation that is mediated through its ability to alter the enhancer epigenetic landscape. We have employed cell-free assays with recombinant chromatin templates and purified factors along with cancer cell analyses to reveal that mutant p53 contributes to enhancer activation by directing H3K4me1 deposition through MLL4 recruitment. Our results define a molecular mechanism for mutant p53-MLL4 interactions in directing epigenetic signatures that are linked to aberrant enhancer and tumor-promoting gene activation.

p53 R273H,P309S and MLL4 colocalize at active enhancers and form physiological associations in colon cancer cells
The recent observation that mutant p53 through direct interactions with NF-B is recruited to a class of active enhancers in a manner that parallels H3K4me1 accumulation in response to chronic TNF␣ signaling (39) raised the question whether mutant p53 regulates the enhancer epigenetic landscape. To begin to address this question, we performed ChIP followed by sequencing (ChIP-Seq) using an antibody specific to MLL4 (KMT2D) in SW480 cells that express p53 with two point mutations, R273H and P309S (hereafter denoted as mutp53) and treated with TNF␣ for 16 h. As demonstrated in Fig. 1A, stringent MLL4 peaks (n ϭ 31,082; false discovery rate Ͻ 0.001) were identified that showed a striking colocalization with mutp53 binding peaks identified in our previously published mutp53 ChIP-Seq data that was also performed in SW480 cells treated with TNF␣ for 16 h (39). Approximately 68% of the MLL4-binding sites show overlap with mutp53 and conversely, 58% of the mutp53-occupied regions overlap with MLL4 peaks in response to chronic TNF␣ signaling (Fig. 1A). To determine whether mutp53 and MLL4 binding overlap at active enhancers, comparative analyses of our MLL4 ChIP-Seq data and our previously published ChIP-Seq data (39) for mutp53 and the enhancer-associated histone marks H3K4me1 and H3K27ac were performed. As shown in the heat maps (Fig. 1B), remarkably, 4,011 active enhancers were identified that are co-occupied by mutp53 and MLL4 in response to chronic TNF␣ signaling. We also compared our previously published global run-on sequencing (GRO-Seq) data in SW480 cells (39) with the MLL4 and mutp53 ChIP-Seq data and found a striking parallel between the mutp53-and MLL4-binding profiles and GRO-Seq signals in response to chronic TNF␣ signaling (Fig. 1B). This strong parallel between mutp53 and MLL4 binding at active enhancers in response to chronic TNF␣ signaling is also demonstrated by the UCSC Genome Browser tracks of mutp53, MLL4, H3K27ac, and H3K4me1 ChIP-Seq data and GRO-Seq signals at the MMP9 and CCL2 enhancers (Fig. 1C), which are among the signal-dependent enhancers that we recently found are bound and activated by mutp53 and NF-B in response to chronic immune signaling (39).
Given the striking overlap of mutp53-and MLL4-binding profiles, we next assessed whether these factors associate with each other under physiological conditions. As shown in Fig. 1D (left), an anti-p53, but not IgG antibody, comparably co-immunoprecipitated mutp53 from SW480 whole cell lysates prepared before and following TNF␣ treatment. In addition, mutp53 was found to co-immunoprecipitate ASH2L and RBBP5 subunits that are shared among the members of the SET1/MLL family and the MLL3/4-specific subunit, UTX (Fig.  1D). Notably, the reciprocal co-immunoprecipitation using an anti-MLL4 antibody also revealed an association of MLL4 with mutp53 and several subunits of the MLL3/4 complex (MLL3/ 4C) that include UTX, ASH2L, and RBBP5 (Fig. 1D, right). These results demonstrate an association between mutp53 and MLL3/4C that is consistent with the TNF␣-induced global overlap of MLL4, mutp53, H3K4me1, H3K27ac, and the enrichment of enhancer-derived transcripts (eRNAs) across the genome.

p53 R273H,P309S regulates enhancer-associated H3K4me1 levels in response to chronic immune signaling
Having identified that mutp53 associates and colocalizes with MLL4 at active enhancers following TNF␣ treatment, we next wanted to investigate whether mutp53 regulates H3K4me1 deposition. To address this question, we performed ChIP-Seq for mutp53 and H3K4me1 in human SW480 colon cancer cells expressing short hairpin RNAs (shRNAs) against p53 and treated with TNF␣ for 16 h. Relative to a nontargeting shRNA against LacZ (control), an shRNA oligonucleotide directed against p53 mRNA significantly reduced mutp53 protein levels both in uninduced and TNF␣-induced cells ( Fig. 2A). The p53 shRNA also significantly decreased MLL2 protein levels, which is consistent with a previous observation that mutp53 regulates the expression levels of the MLL2 gene in breast cancer cells (38). Notably, no change in the protein levels of MLL4 were observed following mutp53 knockdown ( Fig. 2A). In the control cells, composite profiles (Fig. 2B), consistent with the heat maps (Fig. 1B) and the browser tracks shown in Fig. 1C, demonstrate an overlap of the peak binding sites for mutp53 with the peak sites of H3K4me1 enrichment at intergenic regions in response to chronic TNF␣ signaling. Following mutp53 knockdown, a significant loss of mutp53 binding is observed that strikingly parallels with a moderate reduction in H3K4me1 levels (Fig. 2B). In comparison, mutp53 depletion did not affect global H3K4me1 levels as revealed by immunoblot analysis of histone extracts as shown in Fig. 2A, suggesting that the regulation of H3K4me1 deposition by mutp53 is likely to occur in an enhancer-specific manner.
To further investigate the requirement for mutp53 in regulating H3K4me1 levels at specific mutp53 target enhancers, ChIP experiments followed by quantitative PCR (ChIP-qPCR) Mutant p53 co-opts with MLL4 to regulate enhancer activity were performed in uninduced versus TNF␣-induced conditions in control and mutp53 knockdown SW480 cells. As shown in Fig. 2C, low levels of pre-associated mutp53 binding that correlates with low levels of H3K4me1 accumulation were observed before TNF␣ signaling and substantial TNF␣-induced increases in mutp53 binding and H3K4me1 levels were identified at the MMP9 and CCL2 enhancers (amplicon A), but not at the control regions (amplicon B). Under TNF␣-induced conditions, p53 knockdown resulted in a comparable decrease in mutp53 binding (80 and 72%) at the MMP9 and CCL2 enhancers (amplicon A), respectively, as compared with the little to no change at the control regions (amplicon B) (Fig. 2C). Consistent with our ChIP-Seq data (Fig. 2B), the decreased levels of TNF␣-induced mutp53 binding resulted in a significant and comparable reduction in the TNF␣-induced levels of H3K4me1 (70 and 72%) at the MMP9 and CCL2 enhancers,  R273H,P309S and MLL4 in SW480 cells following 16 h TNF␣ treatment. B, heat maps of p53 R273H,P309S and MLL4-binding sites at intergenic regions enriched for p53 R273H,P309S and the histone marks, H3K4me1 and H3K27ac, and rank-ordered by GRO-Seq signal in SW480 cells treated with TNF␣ for 16 h. Each row shows Ϯ 1 kb centered on p53 R273H,P309S peaks. C, UCSC Genome Browser images of p53 R273H,P309S , MLL4, H3K4me1, and H3K27ac ChIP-Seq and GRO-Seq signals at MMP9 and CCL2 gene loci. All ChIP-Seq assays were performed as two biological replicates. D, immunoblot analyses of p53 R273H,P309S (left) and MLL4 (right) co-immunoprecipitation (IP) with lysates prepared from SW480 cells treated with TNF␣ for 0 or 16 h. All co-immunoprecipitations were performed three times, independently.
Mutant p53 co-opts with MLL4 to regulate enhancer activity respectively (Fig. 2C). Under uninduced conditions, we identified little to no effect of mutp53 knockdown on H3K4me1 levels, which is consistent with the significantly lower overall levels of mutp53 and H3K4me1 at the enhancers prior to TNF␣ signaling (Fig. 2C).
We next investigated whether mutp53-dependent changes in H3K4me1 levels are because of mutp53 regulation of MLL4 enhancer binding by performing MLL4 ChIP-qPCR analyses in control versus mutp53 knockdown SW480 cells. Notably, mutp53 knockdown resulted in a significant decrease in TNF␣induced MLL4 binding at the MMP9 and CCL2 enhancers by ϳ45 and 57%, respectively, but not at the control regions ( Fig.  2C), which is consistent with the significant decrease in H3K4me1 levels at these enhancers following the loss of mutp53 enhancer binding. Together, our genome-wide binding data parallel our enhancer-specific analyses, demonstrating that mutp53 directly regulates H3K4me1 levels at specific enhancers in response to chronic TNF␣ signaling.

p53 R273H directly interacts and facilitates primary MLL3/4C recruitment and cooperates with p300 to enhance MLL3/4 binding on reconstituted chromatin
Having identified functional associations between mutp53 and MLL4 at active enhancers in response to chronic TNF␣ signaling in colon cancer cells, we next wanted to investigate the functional relationship between these two factors using

Mutant p53 co-opts with MLL4 to regulate enhancer activity
biochemical assays. First, to determine whether an intact MLL3/4C forms direct interactions with mutant p53, MLL3/4C was purified to near homogeneity through FLAG-tagged PA1, a unique subunit of MLL3/4C (Fig. S1, A-C) using established methods (40). In vitro binding assays revealed that MLL3/4C bound strongly to GST-p53 R273H, but not to GST alone (Fig.  3A). This finding demonstrates a direct p53 R273H-MLL3/4C interaction that is consistent with the physiological association and the functional interplay between these factors at specific enhancers. We further demonstrated the ability of MLL3/4C to interact with GST-tagged p53 WT and the hotspot p53 mutants, R248W and G245S (Fig. S1D). The finding that MLL3/4C interacts with WT p53 is consistent with the findings of a previous study (41).
We next performed in vitro ChIP assays as outlined in Fig. 3B to investigate the effects of p53 R273H on MLL4 binding to recombinant chromatin templates. Chromatin was assembled following established protocols (42) and using recombinant factors (Fig. S1E), reconstituted histone octamers (Fig. S2), and a DNA template that consists of five GAL4-binding sites upstream of the adenovirus (Ad) major late (ML) core promoter and a 390-nucleotide G-less cassette (Fig. 3C). Micrococcal nuclease (MNase) digestion of our chromatinized templates confirmed equivalent nucleosome spacing (Fig. S1F). As depicted in Fig. 3B, chromatin was incubated in the presence or absence of recombinant GAL4 -p53 R273H, the histone acetyl-transferase p300, and MLL3/4C; digested with MNase; and immunoprecipitated with antibodies against p53 R273H or ASH2L, a core subunit of MLL3/4C that is indispensable for its methyltransferase activity (43). The precipitated DNA was analyzed by qPCR using primer sets that scan both the promoter region (amplicon A), containing the GAL4-binding sites, and a distal vector control region (amplicon B) (Fig. 3C). As expected, low levels of p53 R273H binding were identified specifically at the promoter and not the distal control region, which is indicative of the importance of the GAL4-binding sites, which are recognized and bound by GAL4 -p53 R273H (Fig. 3D, column  3). As shown in Fig. 3D (column 4 versus 3), p53 R273H binding to recombinant chromatin was largely unaffected by the presence of MLL3/4C. However, ectopic p300 resulted in a substantial (5-fold) increase in p53 R273H binding at the promoter region (Fig. 3D, column 5 versus 3). Notably, p53 R273H is required to direct MLL3/4C recruitment on chromatin as revealed by the 2-fold increase in ASH2L binding at the promoter (Fig. 3D, column 4 versus 2), which is consistent with our cell-based ChIP data that revealed a requirement for mutp53 in regulating MLL3/4 enhancer interactions (Fig. 2C). In addition, the enhanced binding of ASH2L mediated by p53 R273H was moderately augmented (ϳ1.3-fold) by ectopic p300 (Fig. 3D,  column 7 versus 4). The lack of an effect of p300 alone on ASH2L binding (Fig. 3D, column 6 versus 2) also indicates that the increase in p53 R273H-dependent binding of ASHL2 is Interaction assays were performed three times, independently. B, schematic of the in vitro ChIP assay. C, schematic of the pG 5 ML template indicating the amplicons used for in vitro ChIP-qPCR. D, ChIP-qPCR analyses performed on recombinant chromatin templates using antibodies specific to p53 and ASH2L. Enrichment levels are calculated as -fold enrichment over IgG. Bar graphs represent the average of two independent ChIP assays that are representative of at least three with error bars denoting the S.E.

Mutant p53 co-opts with MLL4 to regulate enhancer activity
likely because of the enhanced p300-mediated stabilization of mutp53 on chromatin (Fig. 3D, column 5 versus 3). Taken together our cell-based and in vitro binding data demonstrate that mutp53-MLL3/4C interactions are important for MLL3/4 recruitment on chromatin and that p53 R273H can act cooperatively with p300 to direct enhanced MLL3/4 binding.

MLL3/4C facilitates p53 R273H-dependent H3K4 monomethylation and enhanced transcription
To further investigate the direct (causal) effects of mutant p53-MLL3/4C interactions on chromatin, we next sought to determine whether p53 R273H impacts MLL3/4C-mediated H3K4me1 deposition by employing histone methyltransferase (HMT) assays as denoted in Fig. 4A. An initial analysis on recombinant histone octamers revealed significant and specific effects of the purified MLL3/4C on H3K4 monomethylation and little to no change in dimethylation and trimethylation (Fig. S2A). On chromatin templates, p53 R273H-dependent increases in H3K4me1 levels were observed that are significantly (4-fold) above the low/negligible levels of H3K4me1 that are observed by MLL3/4C alone (Fig. 4B, lane 3 versus 2). These findings are consistent with the in vitro ChIP data that revealed significant and specific levels of MLL3/4 binding on chromatin templates only in the presence of p53 R273H (Fig. 3D). In addition, the p53 R273H and MLL3/4C-dependent H3K4me1 levels were further enhanced (ϳ6-fold) (Fig. 4B, lane 5 versus lane 3) in the presence of p300, thereby yielding an ϳ30-fold increase in total H3K4me1 levels (Fig. 4B, lane 5 versus 2). Consistent with our HMT assays in which p300 is able to enhance p53 R273H and MLL3/4-mediated H3K4me1, MLL3/4C, albeit less significantly (1.7-fold), enhanced p53 R273H and p300-dependent acetylation (Fig. S2B, lane 5 versus 4). The cooperativity between cofactors, p300, and MLL3/4 on chromatin is consistent with the recently reported identification that MLL4 directly interacts with p300, through the UTX subunit of MLL3/4C, and that these interactions result in the cooperative regulation of H3K4me1 and H3K27ac (44).
To determine whether mutp53, MLL3/4C, and p300-dependent deposition of H3K4me1 and H3K27ac directly contribute to mutp53-dependent transcription, we employed a chromatin-templated cell-free transcription system (42) (Fig. 4, A and  C). A significant level of p53 R273H-dependent transcription was observed that depends on the presence of p300/acetyl-CoA (Fig. 4C, lane 6 versus lane 2). Compared with p300, MLL3/4C alone did not support a detectable level of p53 R273H-dependent transcription (Fig. 4C, lane 7 versus 6). Importantly, however, MLL3/4C enhanced (ϳ2-fold) the p53 R273H-dependent transcription signal in the presence of p300/acetyl-CoA (Fig. 4C, lane 12 versus 6), which is consistent with the significantly higher overall levels of H3K4me1 (Fig. 4B) and H3K27ac (Fig. S2B) that are observed in the presence of MLL3/4 and p300. In contrast, MLL3/4 in the absence of S-adenosylmethionine (SAM) was unable to enhance p53 R273H/p300 -dependent transcriptional activation (Fig. 4C, lane 12 versus 11), which suggests the importance of H3K4me1 in regulating the transcriptional output. Thus, our cell-based findings are corroborated by our in vitro mechanistic studies that demonstrate Mutant p53 co-opts with MLL4 to regulate enhancer activity mutant p53 coordinates H3K4me1 deposition through the chromatin targeting of MLL3/4C, which in turn directs enhanced transcriptional activation.

MLL3/4 regulate the activation of p53 R273H,P309S -bound enhancers in response to chronic TNF signaling
Having established a functional interplay between p53 R273H and MLL3/4C in regulating H3K4me1 deposition and transcriptional activation on recombinant chromatin, we next examined a direct role for MLL3/4C in the regulation of mutp53 enhancers and target genes in response to chronic immune signaling. Given the previously reported functional redundancy of MLL3 and MLL4 (13,14), we generated MLL3/4 double knockdown SW480 cells in which the protein levels of both MLL3 and MLL4 were significantly reduced (Fig. S3A). Under uninduced conditions, MLL3/4 knockdown resulted in little to no effect on MLL4 binding, H3Kme1, or H3K27ac levels at the enhancer regions of MMP9 and CCL2, which is further indicative of the specific binding and roles of MLL4 at mutp53dependent enhancers in response to chronic TNF␣ signaling (Fig. 5A). In TNF␣-treated cells, MLL3/4 knockdown caused substantial losses (53 and 47%) in the TNF␣-induced binding of MLL4 at the enhancer regions of MMP9 and CCL2, respectively, as compared with nonspecific control regions (Fig. 5A). In addition, this loss of MLL4 enhancer binding is accompanied by a notable (ϳ50%) reduction in H3K4me1 levels at both the MMP9 and CCL2 enhancers (Fig. 5A). Although the enhancerspecific H3K4me1 levels were significantly affected by the loss of MLL4 binding, the global H3K4me1 levels remained largely unaffected as revealed by immunoblot analyses of histone extracts prepared from identical cell lines (Fig. S3A). Also, consistent with a role for MLL3/4 in the activation of the MMP9 and CCL2 enhancers, we identified that the TNF␣-induced accumulation of H3K27ac at these enhancers was also significantly reduced (63 and 56%, respectively) following MLL3/4 depletion (Fig. 5A). Moreover, qRT-PCR analyses revealed an ϳ2-fold reduction in the TNF␣-induced expression levels of both the MMP9 and CCL2 eRNAs following MLL3/4 knockdown (Fig. 5B), which also correlated with a decrease in the TNF␣-induced mRNA expression levels of MMP9 and CCL2 (1.8-fold and 11-fold, respectively). Consistent with these changes in enhancer transcription and gene expression, we also identified that MLL3/4 depletion results in a notable (ϳ55%) reduction in the TNF␣-induced binding of RNAPII at the enhancer versus nonspecific control regions of MMP9 and CCL2 (Fig. S3B). This loss of TNF␣-induced RNAPII binding is not because of MLL3/4-dependent stabilization of mutp53 binding because MLL3/4 depletion did not affect the TNF␣induced binding of mutp53 at these enhancers (Fig. S3B). Moreover, the lack of an effect on mutp53 binding in MLL3/4depleted cells is consistent with our in vitro ChIP data that show p53 R273H binding is largely unaffected by the presence of MLL3/4C, whereas p53 R273H is required to regulate MLL3/4C recruitment on reconstituted chromatin.
Having previously established TNF␣and mutp53-dependent alterations in the colon cancer cell transcriptome that are associated with an increase in invasive potential (39), we next wanted to explore whether MLL4-dependent epigenetic alter-ations that are linked to changes in mutp53-dependent gene expression contribute to an increase in cancer cell invasion. As revealed in Fig. 5C, cell invasion assays revealed that TNF␣ treatment results in an ϳ10-fold increase in the number of control cells that pass through the Matrigel. Notably however, MLL3/4 depletion markedly reduced the number of invading cells following TNF␣ signaling by ϳ7-fold (Fig. 5C). Collectively, these findings establish a requirement for MLL4 in regulating TNF␣-induced alterations in enhancer and tumor promoting gene activation that are linked to enhanced cancer cell invasion.

Discussion
In this study, we demonstrated that mutant p53 modulates the chromatin landscape through interactions with MLL3/4. An association between mutant p53 and MLL3/4 was maintained during high-stringency immunoprecipitation experiments and binding assays with recombinant purified proteins. Consistent with the identified mutp53-MLL3/4 interactions, our current study also provides insight into a mechanism in which mutp53 regulates the recruitment of MLL3/4 to chromatin. Specifically, we show that p53 R273H is required for the direct and primary recruitment of MLL3/4C to reconstituted chromatin templates in our cell-free assays. We also establish a requirement for mutp53 in regulating MLL4 recruitment at specific enhancers in response to chronic immune signaling as demonstrated by the significant loss of MLL4 binding following mutp53 depletion. Overall, our results provide strong evidence for a functional relationship between mutp53 and MLL3/4 at specific enhancers following chronic immune signaling. Given that a relatively high number of amino acid residues in the p53 C-terminal domain (CTD) are modified by several posttranslational marks including methylation (45,46), future analysis will address the significance of mutp53-MLL3/4 associations in relation to mutp53 regulation via methylation and/or demethylation mediated by the various subunits of MLL3/4. Such analyses will provide additional insights into mechanisms underlying mutp53-dependent enhancer and gene regulation to advance our understanding of mutp53 GOF properties in human cancer.
Our study highlights an emerging link between WT MLL4 and mutp53 in defective enhancer regulation and the activation of tumor-promoting gene expression. Support for the mechanistic cross-talk between mutp53 and MLL4 at specific enhancers is demonstrated by MLL4 depletion that leads to a loss in the deposition of H3K4me1 and H3K27ac. Under these conditions, concomitant down-regulation of RNAPII binding, eRNA synthesis, and mutp53-dependent gene activation was also observed. In addition, we demonstrate that the enhancer-directed alterations in mutp53 target gene activation mediated by MLL4 and TNF␣ are associated with the biological consequence of increased cancer cell invasion. Whereas our study provides insight into an underlying mechanism for the functional relationship between WT MLL4 and mutp53, future analyses will be directed at unraveling the significance of an association between mutp53 and mutant forms of MLL4. This is supported by previous studies that have demonstrated that MLL4 is among the most recurrently mutated epigenetic Mutant p53 co-opts with MLL4 to regulate enhancer activity enzymes in human cancers including glioblastoma, melanoma, and pancreatic and breast cancers (27,28,47). Characterization of a potential convergence between mutp53 and mutant MLL4 will further advance our understanding of the multidimensional regulation of gene networks in human cancer. It is also noteworthy that we identified a mutp53-dependent regulation of MLL2 as revealed by the significant decrease in MLL2 protein levels in colon cancer cells following mutp53 depletion. This finding is consistent with a previous report showing that mutp53 controls chromatin pathways by regulating the expres-

Mutant p53 co-opts with MLL4 to regulate enhancer activity
sion of the SET/MLL family members, MLL1 and MLL2 in breast cancer (38). These findings, together with our study which shows that mutp53 does not regulate MLL4 protein levels but instead forms functional interactions with MLL4 to regulate enhancer activation, indicate the importance of differential histone methylation mechanisms modulated by mutp53 and the various SET/MLL family members.
A direct mechanistic link between MLL3/4C-mediated methylation, p300-mediated acetylation, and augmented p53 R273H-dependent transcriptional activation was revealed by employing purified proteins, reconstituted chromatin templates, and cell-free assays. We found that although MLL3/4mediated H3K4me1 does not play a primary function in regulating p53 R273H-dependent transcription, this epigenetic change contributes to enhanced transcriptional output in a p53 R273H/p300 -dependent manner. A recent report demonstrated direct interactions between MLL3/4 (UTX subunit) and p300 that promote enhancer activation and in turn, boost transcriptional activation by all-trans-retinoic acid (ATRA) (44). The reciprocal functional interplay between MLL3/4-mediated H3K4me1 and p300-mediated acetylation that is required for enhanced transcriptional activation by mutp53 and ATRA could provide an explanation for the recent reports that demonstrate a significantly reduced effect of catalytically inactive SET domain mutants of MLL3/4 versus MLL3/4 depletion on enhancer transcription (48). It is also noteworthy that the stimulatory effect on p53 R273H/p300 -dependent transcription that is specifically observed for MLL3/4-dependent H3K4me1 versus MLL3/4 in the absence of H3K4me1 (devoid of SAM) provides evidence to support a role of the histone mark H3K4me1 itself in regulating transcriptional activation. A potential role for H3K4me1 in enhancer activation could be mediated through specific H3K4me1 reader proteins that are present in the nuclear extract that is used in our cellfree system. One possible factor could be TIP60, a HAT protein that preferentially binds to H3K4me1 through its chromodomain (49). Further support for this proposed mechanism is also provided by the recent finding that H3K4me1 through the recruitment of cohesin complexes is required to support chromatin interactions at enhancers (50).
In this study, we advance previous analyses of enhancer malfunction in human cancer by revealing a new role for a GOF mutant of the tumor suppressor p53 in shaping the chromatin landscape. Mutp53 and NF-B were previously shown to contribute to each other's recruitment at a subset of cancer cellspecific enhancers that are linked to the activation of a tumorpromoting gene expression program in response to chronic immune signaling (51). Although we demonstrate that p53 WT can also interact with the MLL3/4C, which is consistent with the previously reported cofactor role of MLL3/4C in the regulation of WT p53 tumor suppressor functions (41), it is important to note that the subset of enhancers that were analyzed in this study were previously found to be exclusively bound by mutant and not WT p53 (39). As depicted in the proposed model (Fig. 6), our study demonstrates the significance of mutp53 through interactions with MLL4 in regulating the activation of these mutp5-bound enhancers. The identification of this epigenetic mechanism underlying alterations in enhanc-er-specific histone modifications, enhancer transcription, and the potent induction of mutp53 target genes establishes a new functional link between mutp53 and chromatin pathways in colon cancer cells that are stimulated by chronic TNF␣ signaling.

Cell culture
Human colorectal adenocarcinoma SW480 cells and human embryonic kidney 293T (HEK293T) cells were purchased from American Type Culture Collection (ATCC) and grown in Dulbecco's modified Eagle's medium (DMEM, Gibco Laboratories), supplemented with 10% fetal bovine serum (FBS, Gibco Laboratories). Lentivirus-infected SW480 cells were propagated in DMEM containing 10% FBS and 1.5 g ml Ϫ1 puromycin (Sigma-Aldrich). Stable and inducible SW480 control and p53 knockdown cells that express short hairpins against LacZ or p53 were kindly provided by Xinbin Chen (University of California, Davis), grown in standard DMEM containing 1ϫ penicillin/streptomycin (Gemini Bio-Products) and 1.5 g ml Ϫ1 puromycin, and induced with 1 g ml Ϫ1 doxycycline (Sigma-Aldrich). For experiments with TNF␣ treatment, the indicated cells were treated with 12.5 ng ml Ϫ1 recombinant human TNF␣ (Shenandoah Biotechnology) for 0 or 16 h before harvesting for gene expression or ChIP analyses.

Lentivirus production and transduction
PLKO.1 TRC control and target shRNAs were generated with annealed primers to knockdown MLL3 and MLL4. For lentivirus production, HEK293T cells were transfected using

Mutant p53 co-opts with MLL4 to regulate enhancer activity
Lipofectamine 2000 (Invitrogen) with control or target shRNAs and packaging plasmids psPAX2 and pMD2.G. Medium containing the virus was collected 48 h post transfection and used for infection. SW480 cells were infected with virus-containing medium in the presence of 8 g ml Ϫ1 polybrene (Sigma-Aldrich) for 8 h and subsequently selected with 1.5 g ml Ϫ1 puromycin. Target shRNA sequences are listed in Table S1.

RNA purification and quantitative real-time PCR
Total RNA was isolated from cells using TRIzol LS Reagent (Invitrogen) according to the manufacturer's instructions. cDNA synthesis was performed using ProtoScript II First Strand cDNA Synthesis Kit (New England Biolabs) with random hexamers. qPCR reactions were performed on an Applied Biosystems StepOnePlus Real-Time PCR System using SYBR Green PCR Master Mix (Applied Biosystems) in duplicates and the amplification specificity was examined by melting curve analysis. The relative levels of eRNA and mRNA expression were calculated according to the (⌬⌬Ct) method and normalized to GAPDH. The expression levels determined after TNF␣ treatment are relative to the levels before TNF␣ treatment. Primers for qRT-PCR are listed in Table S2.

Invasion assay
SW480 cells stably expressing control or MLL3/4 shRNAs were pretreated with or without 12.5 ng ml Ϫ1 TNF␣ in low serum media (0.1%) for 24 h prior to plating 2-4 ϫ 10 5 cells from each condition on 24-well PET inserts with 8.0 mm pore size (Falcon), coated with BD Matrigel (BD Bioscience). The lower chambers were filled with high serum media (20%) without TNF␣. Cells that passed through the Matrigel after 24 h were fixed, stained, and counted. Representative scale bars were added using ImageJ.

Immunoblotting
Protein samples were incubated at 95°C for 5 min, resolved by SDS-PAGE, and transferred to PVDF membranes (EMD Millipore) that were subsequently blocked and probed with the indicated antibodies. For MLL3/4 detection, protein samples were resolved on a 3-8% Tris acetate gel (Life Technologies). Reactive bands were detected by ECL (Thermo Scientific Pierce), and exposed to Blue Devil Lite ECL films (Genesee Scientific) for visualization.

Co-immunoprecipitation
Whole cell lysates were prepared from SW480 cells that were treated with 12.5 ng ml Ϫ1 TNF␣ for 0 or 16 h and lysed with the following buffer (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 0.5% Nonidet P-40, 0.5% Triton X-100, 2 mM EDTA), supplemented with protease inhibitor cocktail (PIC) (Sigma) on ice for 30 min. Cleared lysates were incubated with the indicated antibodies overnight at 4°C, followed by incubation with protein A Dynabeads (Sigma-Aldrich) for 2 h at 4°C. Beads were subsequently washed with lysis buffer three times prior to elution and analysis of protein complexes by immunoblotting.

Recombinant protein purification and in vitro binding assays
GSH S-transferase-tagged p53 WT, R175H, R273H, R248W, and G245S were expressed in Escherichia coli Rosetta cells and purified using GSH Sepharose 4B (GE Healthcare Life Sciences). For in vitro binding assays, the MLL3/4 complex was precleared with GSH Sepharose beads for 30 min at 4°C, and then incubated with 1.5 g of the GST fusion proteins overnight at 4°C in binding buffer (25 mM HEPES-KOH, pH 7.8, 150 mM NaCl, 0.1% Nonidet P-40, 1.5 mM MgCl 2 , 10% glycerol). The bound proteins were washed three times with binding buffer, eluted, and analyzed by immunoblotting.

Chromatin immunoprecipitation and ChIP-Seq
SW480 cells were treated with i) 12.5 ng ml Ϫ1 TNF␣ for 0 or 16 h or ii) indicated shRNAs and subsequently exposed to 12.5 ng ml Ϫ1 TNF␣ for 0 or 16 h. Cells were cross-linked in 6 mM disuccinimidyl glutarate (DSG) (ProteoChem) in PBS for 30 min at room temperature followed by the addition of 1% final formaldehyde concentration for 10 min at room temperature. Cross-linking was quenched with 125 mM glycine (Fisher Scientific). Cell pellets were resuspended in lysis buffer (20 mM Tris-HCl, pH 7.5, 300 mM NaCl, 2 mM EDTA, 0.5% Nonidet P-40, 1% Triton X-100, 1 mM PMSF, PIC) and incubated on ice for 30 min, followed by homogenization. Nuclear pellets were collected and resuspended in shearing buffer (0.1% SDS, 0.5% N-lauroylsarcosine, 1% Triton X-100, 10 mM Tris-HCl, pH 8.1, 100 mM NaCl, 1 mM EDTA, 1 mM PMSF, PIC), and chromatin was fragmented to an average size of 200 -500 bp with a Bioruptor Pico (Diagenode). Precleared chromatin was immunoprecipitated with the indicated antibodies overnight at 4°C and Mutant p53 co-opts with MLL4 to regulate enhancer activity immunocomplexes were collected with protein A Dynabeads. The immunocomplexes were washed eight times in wash buffer (50 mM HEPES-KOH, pH 7.6, 500 mM LiCl, 1 mM EDTA, 1% Nonidet P-40, 0.7% sodium deoxycholate, 1 mM PMSF, PIC), followed by two TE washes, and subsequently eluted in elution buffer (50 mM Tris-HCl, pH 8.0, 10 mM EDTA, 1% SDS). The eluates were incubated overnight at 65°C to reverse the crosslinks and ChIP DNA was purified using DNA Clean & Concentrator Kit (Zymo Research) according to the manufacturer's instructions before qPCR analysis on an Applied Biosystems StepOnePlus Real-Time PCR system using SYBR Green PCR Master Mix in duplicates. The relative amounts of ChIP DNA were quantified relative to corresponding input samples. Primers for ChIP-qPCR are listed in Table S3.
For ChIP-Seq experiments, the IPs were performed as described above. The eluted ChIP DNA was quantitated using a Qubit 2.0 Fluorometer (Invitrogen), and 1-5 ng of ChIP DNA was used to prepare the sequencing libraries from at least two biological replicates using the ChIP TruSeq Sample Prep Kit according to the manufacturer's instructions (Illumina). Briefly, ChIP DNA samples were end-repaired and adaptors were ligated to the ends of the DNA fragments. Adaptor-ligated ChIP DNA fragments with average size of 350 bp were used to construct libraries and single-end sequenced (75 bp) on Illumina HiSeq 4000. Reads were mapped to hg38 human genome using Bowtie2 software 38 and default parameters. The mapped reads were then processed to make Tag Directory module using HOMER for filtering. Briefly, PCR duplications were removed and only uniquely mapped reads were kept for further analysis. Enrichment for p53 R273H,P309S and H3K4me1 deposition were called using findPeaks module from HOMER using preset parameters for transcription factors and histones, respectively, and compared with the input samples. The genome browser files for the resulting reads were generated by using make UCSCfile module from HOMER. Deeptools were used to generate heat maps.

Recombinant chromatin assembly, in vitro ChIP, HAT, and HMT assays
In vitro chromatin assembly was performed as previously described (42) using a pG 5 ML DNA template containing 5ϫ GAL4 response elements and reconstituted histone H3 octamers, ACF1, ISWI, and NAP1. ChIP, HAT, and HMT assays were performed in two consecutive steps: i) 150 ng of chromatin, 100 ng of GAL4-p53 R273H, 30 ng of p300, and 20 M acetyl CoA were incubated at 30°C for 30 min, 30°C for 30 min in reaction buffer (50 mM HEPES, pH 7.8, 30 mM KCl, 0.25 mM EDTA, 5.0 mM MgCl2 , 5.0 mM sodium butyrate, 2.5 mM DTT), followed by ii) addition of 1 l of the MLL3/4C and 20 M SAM for an additional incubation at 30°C for 60 min. For ChIP assays, reactions were terminated on ice followed by MNase (Sigma-Aldrich) digestion in buffer R (10 mM HEPES, pH 7.9, 10 mM KCl, 0.5 mM EGTA, 10% glycerol) at room temperature for 10 min. Digestion reactions were then quenched with the addition of 20 mM EDTA and subsequently immunoprecipitated with the indicated antibodies overnight at 4°C. Immunocomplexes were washed, eluted, and processed for analysis by qPCR as described above. For HAT and HMT assays, reactions were resolved on 15% SDS-PAGE gels and analyzed by immunoblotting.

In vitro chromatin-templated transcription assays
Chromatin-templated transcription was performed as described (42) with minor modifications. Transcription steps are as follows: 1) activator binding: GAL4-p53 R273H (100 ng) was incubated with recombinant chromatin template (40 ng) in reaction buffer at 30°C for 20 min; 2) chromatin-based histone acetylation: p300 (25 ng) and acetyl CoA (20 M) were added to the reactions and incubated at 30°C for 30 min; 3) chromatinbased histone methylation: MLL3/4C and SAM (20 M) were added to the reactions and incubated at 30°C for 60 min; 4) transcription: 5.0 l (10 mg ml Ϫ1 ) of HeLa nuclear extract and nucleotide mixture with 10 Ci of [␣-32 P]CTP (3000 Ci/mmol, PerkinElmer Life Sciences) were sequentially added to the reaction for 20 min and 50 min at 30°C, respectively. The radiolabeled RNA was purified and resolved on 5% polyacrylamide (37.5:1) with 8.0 M urea and analyzed by autoradiography.

Data availability
All sequencing data that support the findings of this study have been deposited in the National Center for Biotechnology Information Gene Expression Omnibus (GEO) and are accessible through the GEO Series Accession numbers GSE102796 and GSE115985. All other relevant data are available from the corresponding author upon reasonable request.